HAL Id: hal-02344585https://hal.archives-ouvertes.fr/hal-02344585

Submitted on 4 Nov 2019

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Involvement of the GABAA receptor α subunit in themode of action of etifoxine

César Mattei, Antoine Taly, Zineb Soualah, Ophélie Saulais, Daniel Henrion,Nathalie C. Guérineau, Marc Verleye, Christian Legros

To cite this version:César Mattei, Antoine Taly, Zineb Soualah, Ophélie Saulais, Daniel Henrion, et al.. Involvement of theGABAA receptor α subunit in the mode of action of etifoxine. Pharmacological Research, Elsevier,2019, 145, pp.104250. �10.1016/j.phrs.2019.04.034�. �hal-02344585�

Page1of37

Involvement of the GABAA receptor a subunit in the mode of action of 1

etifoxine 2

3

César Matteia*, Antoine Talyb, Zineb Soualaha, Ophélie Saulaisa, Daniel Henriona, Nathalie C. 4

Guérineaua,c, Marc Verleyed and Christian Legrosa* 5

6

aInstitut MITOVASC, UMR CNRS 6015 - UMR INSERM U1083, Université d’Angers, 3 Rue Roger 7

Amsler 49100 ANGERS, France 8

bTheoretical Biochemistry Laboratory, Institute of Physico-Chemical Biology, CNRS UPR9080, 9

University of Paris Diderot Sorbonne Paris Cité, 75005 Paris, France 10

cpresent address: Institut de Génomique Fonctionnelle, CNRS UMR5203; INSERM U1191; 11

Université Montpellier, 141 rue de la Cardonille, 34094 Montpellier CEDEX 05 12

dBiocodex, Department of Pharmacology, Zac de Mercières, 60200 Compiègne, France 13

14

* Co-corresponding authors. 15

E-mail addresses: christian.legros@univ-angers.fr (C. Legros); cesar.mattei@univ-angers.fr (C. 16

Mattei) 17

18

19

20

21

22

23

Page2of37

ABSTRACT 24

Etifoxine (EFX) is a non-benzodiazepine psychoactive drug which exhibits anxiolytic effects through 25

a dual mechanism, by directly binding to GABAA receptors (GABAARs) and to the mitochondrial 18-26

kDa translocator protein, resulting in the potentiation of the GABAergic function. The b subunit 27

subtype plays a key role in the EFX-GABAAR interaction, however this does not explain the anxiolytic 28

effects of this drug. Here, we combined behavioral and electrophysiological experiments to challenge 29

the role of the GABAAR a subunit in the EFX mode of action. After single administrations of 30

anxiolytic doses (25-50 mg/kg, intraperitoneal), EFX did not induce any neurological nor locomotor 31

impairments, unlike the benzodiazepine bromazepam (0.5-1 mg/kg, intraperitoneal). We established 32

the EFX pharmacological profile on heteropentameric GABAARs constructed with α1 to α6 subunit 33

expressed in Xenopus oocyte. Unlike what is known for benzodiazepines, neither the γ nor δ subunits 34

influenced EFX-mediated potentiation of GABA-evoked currents. EFX acted first as a partial agonist 35

on α2b3γ2S, α3b3γ2S, α6b3γ2S and α6b3d GABAARs, but not on α1b3γ2S, α4b3γ2S, α4b3d nor 36

α5b3γ2S GABAARs. Moreover, EFX exhibited much higher positive allosteric modulation towards 37

α2b3γ2S, α3b3γ2S and α6b3γ2S than for α1b3γ2S, α4b3γ2S and α5b3γ2S GABAARs. At 20 µM, 38

corresponding to brain concentration at anxiolytic doses, EFX increased GABA potency to the highest 39

extent for α3b3γ2S GABAARs. We built a docking model of EFX on α3β3γ2S GABAARs, which is 40

consistent with a binding site located between α and β subunits in the extracellular domain. In 41

conclusion, EFX preferentially potentiates α2b3γ2S and α3b3γ2S GABAARs, which might support 42

its advantageous anxiolytic/sedative balance. 43

44

Chemical compounds studied in this article: etifoxine (PubChem CID: 171544), bromazepam 45

(PubChem CID: 2441), diazepam (PubChem CID: 3016) 46

Page3of37

Keywords: etifoxine; GABAA receptors; a subunit; anxiolysis; behavioral pharmacology; EFX-47

binding mode 48

49

50

Highlights 51

• We investigated the influence of a subunits of GABAAR on the mode of action of etifoxine, 52

a non-benzodiazepine compound. 53

• Etifoxine induces anxiolysis without locomotion impairment and sedation in mice. 54

• Etifoxine strongly potentiates α3b3γ2S and moderately α2b3γ2S and α6b3γ2S GABAARs 55

compared to other GABAARs. 56

• A docking model of EFX with α3b3γ2S GABAAR reveals a binding site at the α/b interface, 57

close to the GABA-binding pocket. 58

59

60

61

Abbreviations 62

a(1-6)GABAARs, a1 to a6 subunit-containing GABAA receptors; BZD, benzodiazepine; BZP, 63

bromazepam; DZP, diazepam; EFX, etifoxine; GABAAR, GABAA receptors; PAM, positive allosteric 64

modulator; TEVC, two-electrode voltage-clamp 65

66

Page4of37

1. Introduction 67

GABAARs are heteropentameric membrane proteins that belong to the cys-loop ligand-gated 68

ion channel superfamily [1]. They are permeant to chloride ions in response to GABA and decrease 69

neuronal excitability through membrane hyperpolarization. To date, 19 mammalian GABAAR 70

subunits have been described and cloned (α1-6, β1-3, γ1-3, δ, ε, π, θ, ρ1-3) [1]. The putative 71

combination of these subunits provides a large heterogeneity of GABAARs, with a stoichiometry of 72

2α, 2β and a complementary subunit (mainly γ or δ). The contribution of GABAARs to fast or slow 73

neuronal inhibition depends on their stoichiometry, their tissue distribution and their synaptic or 74

extrasynaptic location. The most frequent assembly of synaptic receptors is α(1-3)β(1-3)γ(1-3), 75

whereas extrasynaptic receptors dominantly contain α4 or α6 with β(1-3) and δ, or α5, β(1-3) and 76

γ2 [1,2]. These differences in stoichiometry and distribution support their different 77

neurophysiological functions and pharmacological properties [3,4]. 78

GABAARs are targeted by benzodiazepines (BZDs) and other drugs for the treatment of 79

anxiety, epilepsy and sleep disorders [5,6]. BZDs act as positive allosteric modulators (PAMs) of 80

GABAARs by binding to a site at the interface between g2 subunit and α subunits [7]. Classical BZDs, 81

such as diazepam (DZP, Fig. 1), bromazepam (BZP, Fig. 1) and lorazepam, exhibit similar 82

pharmacological profile in behavioral tests [8,9] and display poor selectivity over GABAARs which 83

contain α1, or α2 or α3 or α5 (α1GABAAR, α2GABAAR, α3GABAAR, or α5GABAAR) [10], which 84

explains their undesirable effects, including withdrawal symptoms, sedation, amnesia, cognitive 85

impairments and aggressiveness. Indeed, α1GABAARs are associated with sedation, BZD addiction, 86

anterograde amnesia, anticonvulsant activity and cortical plasticity [10-12]. α2GABAARs and 87

α3GABAARs have been linked to anxiolysis, antihyperalgesia and myorelaxation [13-15]. 88

α5GABAARs are believed to be correlated to sedation, cognitive impairments and more recently, 89

anxiolysis [14-17]. 90

Page5of37

Etifoxine (2-ethylamino-6-chloro-4-methyl-4-phenyl-4H-3,1-benzoxazine hydrochloride, 91

EFX, Fig. 1) is a non-BZD compound that exhibits anxiolytic and anticonvulsant effects in rodents 92

[18] and is used for the treatment of anxiety-related disorders in humans [19,20]. EFX also displays 93

anti-hyperalgesic and anti-inflammatory properties in different animal models [21,22]. Both in vitro 94

and in vivo studies in rats suggested that the anxiolytic effects of EFX involve a dual mechanism of 95

action, by directly binding to central GABAARs and to the mitochondrial 18-kDa translocator protein 96

(TSPO) with, as a result, potentiation of the GABAergic function [23,24]. Indeed, it has been shown 97

that EFX activates TSPO through a direct binding and consecutively stimulates the synthesis of 98

neurosteroids, such as, allopregnanolone, which act as PAMs of GABAARs [25-27]. Although the 99

affinity of EFX for GABAARs was twice higher than the one for TSPO (Ki of 6.1 µM vs Ki of 12.7 100

µM), the predominance of one of the effect over the other, i.e. direct GABAARs binding or through 101

TSPO activation, in mediating its anxiolytic effect, is still debated [25,28,29]. 102

The importance of the β subunit in the mode of action of EFX on GABAARs has been clearly 103

evidenced [24]. Constitutively-open homopentameric βGABAARs are inhibited by EFX. In addition, 104

α1GABAARs and α2GABAARs embedding β2 or β3 are more sensitive to EFX than α1GABAARs and 105

α2GABAARs with β1. These data underline the importance of the β subunit in the EFX-GABAAR 106

interaction and for EFX-potentiation of GABA-induced currents of heteromeric GABAARs. However, 107

homopentameric β2GABAARs are less sensitive to EFX than homopentameric β1GABAARs [24], 108

suggesting that the nature of the α subunit might also play a role in EFX-GABAARs interaction. In 109

addition, at anxiolytic doses, EFX has no sedative effects nor locomotion impairment in humans [19] 110

or rodents [30] and this could hardly be explained by the equal potency of EFX on α2GABAAR and 111

α1GABAAR since the latter is associated with sedation [10,12]. Thus, we hypothesized that the 112

pharmacological profile of EFX reflects different sensitivities of all α subtypes containing-GABAARs. 113

In this study, we first compared EFX and BZP in anxiolysis, sedation and locomotor 114

impairment behavioral tests in acute conditions, in mice. The pharmacological effects of both EFX 115

Page6of37

and BZP already appear after a single administration [8,9,29,30]. Here, we determined the anxiolytic 116

doses of EFX and their possible influences on motor performance and arousal. We then assessed the 117

impact of α subunit isoforms on the effects of EFX on GABA-evoked currents. We characterized the 118

pharmacological profile of EFX on murine synaptic GABAARs (α1b3γ2S, α2b3γ2S, α3b3γ2S, 119

α4b3γ2S and α6b3γ2S) and extrasynaptic GABAARs (α5β3γ2S, α4β3δ and α6β3δ) using 120

electrophysiology. This pharmacological study was completed with a 3D model, showing the 121

interaction between EFX and GABAARs. Our results demonstrate that the EFX mode of action 122

involves both α and β, but not γ or δ, subunits. 123

124

Page7of37

2. Materials and methods 125

2.1 Ethical statements 126

All animal procedures were carried out in accordance with the European Community council 127

directive 2010/63/EU for the care and use of laboratory animals and were approved by our respective 128

local ethical committees (N°CEEA.45 and N°CEEA.72 for mice and N°CEEA.2012.68 for Xenopus, 129

https://www.ceea-paysdelaloire.com/) in addition to the French Ministry of Agriculture 130

(authorization N°B49071 and N° 02200.02). The NC3R's ARRIVE guidelines were followed in the 131

conduct and reporting of all experiments using animals. 132

2.2 Animal care and conditioning 133

Experiments were carried out using 7- to 9-week-old Balb/cByJ mice (25-30 g) purchased 134

from Charles River Laboratory (Les Oncins, France). Ten mice per translucent polypropylene cage 135

(internal dimensions in mm: 375 x 375 x 180, L x W x H) were housed under standard laboratory 136

conditions (22 ± 2° C, 12-h light/dark cycle, lights on at 7:00 AM) with food (AO4, SAFE, France) 137

and tap water available ad libitum. No less than one week of rest followed their arrival. Mice were 138

habituated to the testing room at least 60 min before performing any behavioral evaluation. All tests 139

occurred between 9:00 AM and 3:00 PM. The behavioral tests were performed by two well-trained 140

experimenters, who remained unaware of the administered treatment. In addition, all equipment was 141

wiped with 70% ethanol between animals to erase the olfactory stimuli. All experiments were 142

performed in a randomized manner. Single administrations of EFX (12.5-150 mg/kg, expressed as 143

hydrochloride salt) or BZP (0.25-1 mg/kg) were given by the intraperitoneal (IP) route, 30 min before 144

each test, except in the stress-induced hyperthermia test in which the compounds were administered 145

60 min before the test. Studies have shown that both compounds have a similar profile with plasma 146

peak at 15-30 min [25,31-33]. The control animals received an equivalent volume of vehicle (0.9% 147

NaCl, 1% tween 80 (v/v)). One male C57Bl/6N mouse was used for the cloning experiments. This 148

Page8of37

mouse had been included in a control group (not treated) from a previous protocol in which mice 149

were purchased from Janvier Labs (Le Genest-Saint-Isle, France). Euthanasia was performed using 150

CO2 (3 ml/min, 4 min). Adult female Xenopus laevis were purchased from CRB (Rennes, France) and 151

had been bred in the laboratory in strict accordance with the recommendations of the Guide for the 152

Care and Use of Laboratory Animals of the European Community. Oocytes were harvested from 153

mature female Xenopus laevis frogs under 0.15% tricaine anaesthesia. All animals recovered after 2-154

3 h. Every female is operated every three months, not less. A single female was used no more than 5 155

times. 156

2. 3 Compounds 157

EFX hydrochloride (batch 653, Biocodex, France), and BZP (batch 5788, Francochim, 158

France) (Fig. 1) were suspended in vehicle and administered (IP) in a volume of 10 ml/kg of body 159

weight. For electrophysiology, EFX and DZP (batch 105F0451, Sigma, France) were dissolved in 160

dimethylsulfoxide (DMSO) resulting in a maximal concentration of 0.1% DMSO, for oocyte 161

perfusion (control experiments were performed to demonstrate no effect of DMSO). All other 162

reagents and solvents were obtained from Sigma-Aldrich Merck (Saint-Louis, MO, USA) or Thermo 163

Fisher Scientific (Waltham, Massachusetts, USA). 164

165

166

167

168

169

170

Fig. 1. Structure of the positive allosteric modulators of GABAARs used in this study. The 2D structure of EFX (PubChem 171 CID: 171544), DZP (PubChem CID: 3016) and BZP (PubChem CID: 2441) are illustrated. 172

173

Etifoxine (EFX) Diazepam (DZP) Bromazepam (BZP)

Page9of37

2.4 Anxiety-related behavioral assessment 174

Stress-induced hyperthermia (SIH) is defined as the increase in body temperature observed 175

when a subject is exposed to an external stressor [34]. The day preceding the experiment, the animals 176

were isolated in smaller cages (dimensions: 265x160x140 mm). The body temperature (± 0.1°C) of 177

one singly housed mouse was measured twice via a rectal probe (YSI n°423; 2 mm diameter) coupled 178

to a thermometer (Letica-Temp812 model-Italia) at an interval of 10 min. The rectal temperature 179

measurement procedure (handling, insertion of the probe) constitutes the stressor. Drugs were injected 180

60 min before the first measurement (T1), followed by a second temperature measurement 10 min 181

later (T2). Methodological experiments have shown that the optimal conditions for drug testing are 182

found with an injection-test interval of 60 min or longer to avoid residual effects of the injection 183

procedure. Indeed, using a 60 min-injection-stress interval results in a hyperthermic response 184

comparable to animals that are not injected [35,36]. The reduction of SIH (∆T = T2-T1) is considered 185

to reflect an anxiolytic-like effect [34-36]. Defensive burying after novel object exploration is a 186

behavior that can be elicited in rodents in response to aversive or new stimuli [37]. Mice were singly 187

housed in smaller cages (see above) with a sawdust depth of 2.5 cm the day before the test. Each 188

mouse was confronted with an unfamiliar object (4x4x6 cm; aluminium) introduced in the centre of 189

the cage. The time of contact or exploring the object (snout pointing toward the object at a distance 190

< 1 cm with or without burying) was recorded for 10 min. The object was cleaned with alcohol (10%) 191

between each trial. This behavioral approach reveals an anxiety-like or fear state and its suppression 192

is associated with a reduction of the anxiety-like behavior [37]. 193

2.5 Spontaneous locomotor activity 194

Testing was conducted in a quiet room under a light level of approximately 400 lux. The motor 195

activity cages (dimensions: 265x160x140 mm) were made of clear plastic and were changed between 196

each animal; these cages contained a minimum amount of sawdust. The locomotor activity was 197

Page10of37

measured by infra-red beam interruptions that were counted by a control unit (OptaVarimex, 198

Columbus, Ohio, USA). The sensitivity of this unit was set so that walking (horizontal activity) and 199

rearing (vertical activity) were measured. The beam breaks corresponding to spontaneous locomotor 200

activity were measured for 15 min. 201

2.6 Rotarod performance 202

The rotarod test assesses motor performance by measuring the capacity of mice to remain on 203

a 3-cm-diameter rod revolving 16 rpm (model 7600; Ugo Basile, Comerio, Italy). The mice were 204

trained to walk on the rotarod until they could complete three consecutive 120 s sessions without 205

falling off the rod. Twenty-four hours later, selected animals were treated with drugs before being 206

challenged. The rotarod performance time was measured three times, up to 120 s, and the mean was 207

adopted as the performance time for each animal. 208

2.7 GABAAR subunit cDNA expression vectors 209

pRK7 plasmids containing cDNAs encoding rat α2 subunits were a kind gift from Professor 210

Harmut Lüddens (University of Mainz, Mainz, Germany). The GABAAR α2 subunits from rat and 211

mouse are identical in amino acid sequence. pGW1 (=pRK5) plasmids containing cDNAs encoding 212

mouse β3 and γ2S subunits were kindly provided by Professor Steven J. Moss (Tufts University, 213

Boston, USA). The cDNAs encoding the α3, α4, α5, α6 and d subunits used in this work were cloned 214

in mouse brain as described below. 215

2.8 RNA extraction, RT-PCR and cloning of full-length cDNAs encoding α3-6 and d subunits 216

The brain was dissected from a male C57Bl/6N mouse for RNA extraction and purification. 217

Total RNA was then extracted using TRIzol® Reagent (Ozyme/Biogentex, France). First strand 218

cDNAs were synthesised from 5 µg of total RNA using SuperScriptTM III First-Strand Synthesis 219

System Super Mix (Invitrogen, USA) in the presence of oligo (dT)20, according to the manufacturer’s 220

Page11of37

instructions. cDNAs encoding α3-6 and d subunits were amplified using gene-specific primer pairs 221

encompassing each ORF (Supplemental Table 1) and high-fidelity thermostable DNA polymerase 222

(Advantage 2 Proofreading Polymerase kit, Clontech, Saint-Germain-en-Laye, France). cDNAs 223

fragments were purified with the Nucleospin PCR Cleanup Kit (Macherey-Nagel, Hoerdt, Germany) 224

and were subsequently cloned into PCR® 4 TOPO® (Invitrogen). Each clone was sequenced twice 225

on both strands using universal sense and reverse primers by GATC Biotech (Konstanz, Germany). 226

Sequence analyses were performed using BioEdit sequence analysis software. To transfer α3-6 and d 227

subunits ORFs into the pRK5 expression vector, we adapted the ligase-free method for directional 228

cloning [38]. Plasmids and cDNA inserts were separately prepared by PCR using the proof reading 229

polymerase, KOD DNA polymerase (Merck Millipore, Fontenay sous Bois, France). To generate 230

sticky-end cDNAs, two individual PCR reactions were performed, PCR1 and PCR2, with gene-231

specific primers containing short overhangs that allow annealing with the complementary overhangs 232

of the plasmid. pRK5 plasmid was modified to clone the GABAAR subunit ORFs flanked at their 5’ 233

end by alfalfa mosaic virus (AMV) coat protein (RNA 4) and at their 3’end by 3ʹ -untranslated 234

regions (UTRs) from the Xenopus β-globin gene (3UTRXBG). The combination of both UTRs has 235

been shown to improve expression in both oocytes and mammalian cells [39]. First, a fusion of AMV 236

and 3’UTRXBG was constructed and cloned into pRK5 between EcoRI and XbaI. The resulting 237

modified vector (pRK5-5AMV-3UTRXBG) was used as a template for two individual PCRs with the 238

following pair primers: 5’-TAAACCAGCCTCAAGAACACCCGA-3’ with 5’ 239

GGTGGAAGTATTTGAAAGAAAATTAAAAATA-3’ (PCR1), and 5’-240

AAGCTTGATCTGGTTACCACTAAACC-3’ with 5’-241

AAAATTAAAAATAAAAACGAATTCAATCGATA-3’ (PCR2). PCR1 and PCR2 products were 242

purified and mixed in T4 ligase buffer. To generate cDNA with sticky ends, the amplicons were 243

subjected to melting and reannealing, as previously described [38]. Inserts containing GABAAR 244

subunit ORFs were also prepared in two individual PCRs (LIC-PCR1 and 2) with gene-specific 245

Page12of37

primer pairs (see Table S1 in supplementary data). Sticky-end inserts were obtained as described for 246

the plasmid preparations. For each construct, sticky-ends plasmid and GABAA cDNA preparation 247

were assembled in T4 ligase buffer and incubated for 2 h at 22°C. One to two microliters of this 248

assemblage were used to transform chemically competent E. coli cells (DG1, Eurogentec, Seraing, 249

Belgium). The resulting clones were sequenced as described above. 250

2.9 Expression of GABAARs in Xenopus oocytes 251

Adult female Xenopus laevis (CRB, Rennes, France) were anaesthetized in ice-cold water 252

with 0.15% Tricaine (3-aminobenzoic acid ethyl ester, Sigma). Ovarian lobes were collected and 253

washed in standard oocyte saline (SOS containing 100 mM NaCl, 2 mM KCl, 1.8 mM CaCl2, 1 mM 254

MgCl2, 5 mM HEPES, pH 7.4). Stage V-VI oocytes were partially defolliculated by enzymatic 255

treatment with 2 mg/ml collagenase (type IA, Sigma) in Ca2+-free SOS for 60 min. To express 256

functional GABAAR, cDNA mixtures were directly injected into the nucleus (animal pole) of 257

individual defolliculated oocytes in different volumes of DNA solution at a concentration of 50 ng/µl 258

using a nanoinjector (Drummond Nanoject) (see Table S2 for receptor stoichiometry and DNA 259

quantity injected). Following injection, the oocytes were kept at 18°C in SOS supplemented with 260

gentamycin (50 μg/ml), penicillin (100 UI/ml), streptomycin (100 μg/ml), and sodium pyruvate 261

(2.5 mM). The incubation medium was replaced every two days. Oocytes were incubated 1 to 2 days 262

after DNA injection, depending on the GABAAR subtype. 263

2.10 Electrophysiological recordings 264

Injected oocytes were tested for GABAAR expression, at a holding potential of -60 mV using 265

a two-electrode voltage clamp amplifier (TEV-200, Dagan Corporation, Minneapolis, USA). Digidata 266

1440A interface (Axon CNS Molecular Devices, California, USA) and pCLAMP 10 software (Axon 267

CNS Molecular Devices) were used for acquisition. Cells were continuously superfused with standard 268

oocyte saline (SOS) at room temperature and were challenged with drugs in SOS. Electrodes were 269

Page13of37

filled with 1 M KCl / 2 M K acetate and had typical resistances of 0.5–2 MΩ in SOS. Drugs were 270

perfused at a flow rate close to ~4 ml/min. EFX and DZP were applied for 2 min before co-application 271

of GABA at EC10, (determined for each GABAAR subtype, see Table S2 in supplementary data) until 272

the current response peaked. EFX was tested at concentrations ranging from 2 to 100 μM, 273

corresponding to its clinical use. Between the two applications, the oocytes were washed in SOS for 274

10-15 min to ensure full recovery from receptor desensitization (see Fig. 4A inbox). To control 275

whether GABA-evoked currents were mediated by ternary α1-6β3γ2SGABAARs, control 276

experiments were performed using SOS containing 10 µM Zn2+ to inhibit binary GABAARs [40]. Data 277

were analysed using pCLAMP 10 software. Data are expressed as the mean ± SEM of 6-10 oocytes 278

generated from at least two collections. Concentration-effect relationships were analysed using the 279

following equation: Y=Ymin + (Ymax-Ymin)/(1+10((LogEC50-X).nH)), where X is the concentration of EFX, 280

Ymin and Ymax are the minimum and highest responses, and nH is the Hill coefficient. 281

2.11 Molecular model design 282

The template chosen for homology modelling was the recently solved structure of a GABAAR 283

(pdb code 4COF). The sequences of the human α2, β3 and γ2S GABAAR subunits were aligned with 284

those of the template (β3) using T-Coffee software [41]. The model was then prepared by homology 285

modeling using Modeler version 9.5 software [42] with default settings. One hundred models were 286

prepared, and the best model, according to the Discrete Optimized Protein Energy function (DOPE), 287

was selected. Side chains in the models were improved with Scwrl4 [43]. The whole model was then 288

improved with CHARMM [44,45]. Disulfide bridges formed between neighbouring cysteines both in 289

the ‘Cys-loop’ and between the M1 and M3 transmembrane helices in α and γ subunits, as recently 290

proposed [45]. The model was then subjected to minimization with decreasing harmonic potential. 291

2.12 Docking 292

Page14of37

The docking had been performed with AutoDock Vina [46]. The ligands and proteins were 293

prepared with prepare_ligand4.py and prepare_receptor4.py scripts, respectively. The side chains of 294

amino-acids in the binding site were made flexible (a3: Ser257 Ser258 and β3 Gln66, Tyr87, Gln89, 295

Tyr91). Each ligand was docked 100 times in a large cube of 30 Å in each dimension. Fig. 7 was 296

prepared with PyMOL (DeLano W.L. (2010) The PyMOL Molecular Graphics System, version 1.6, 297

Schrodinger, LLC, New York). 298

2.13 Data analysis and statistics 299

Data are presented as the mean ± SEM. Behavioral data were analysed by one-way ANOVA 300

followed by Dunnett’s post-hoc test for comparison with the vehicle group. In cases in which the two 301

conditions (normality of the data and equality of variances) were not fulfilled, the non-parametric 302

Kruskal-Wallis procedure was used, followed by the post-hoc Dunn’s test to evaluate the statistical 303

significance between the vehicle and treated groups. All test analyses were carried out by observers 304

who were blinded to the experimental procedures. Sample sizes (number of animals in the behavioral 305

studies) were not predetermined by a statistical method. Each behavioral experiment group included 306

at least 10 animals and this sample size needed to detect significant effects was based on experience 307

from previous studies. Significance tests between groups in the electrophysiological studies were 308

performed using variance analysis (one-way ANOVA) followed by Tukey’s post-hoc test for 309

comparison of all groups or the non-parametric Mann and Whitney test when appropriate. Concerning 310

the electrophysiological experiments, we compiled data from different batches of oocytes and we 311

excluded data, in case of potential drift (> 0.6 mV) after pulling out the electrodes from the oocytes 312

and when current amplitudes were <10 nA or > 2 µA. GraphPad Prism 7.02 (GraphPad Software, San 313

Diego, USA) was used for all graphs and statistical analyses. Differences with p<0.05 were 314

considered significant (* for p<0.05, ** for p<0.01, *** for p<0.001, **** for p<0.0001). 315

316

Page15of37

3. Results 317

3.1 Anxiolytic effects of EFX 318

Previous studies have shown that EFX exhibits anxiolytic effects using conventional 319

behavioral tests such as elevated plus maze and dark-light box tests [33,47]. Here, we evaluated the 320

effect of EFX on stress and anxiety-related behaviors (stress-induced hyperthermia and novel object 321

exploration) to determine and confirm its anxiolytic doses in comparison with BZP. In non-treated 322

animals, handling stress resulted in a rise in body temperature close to 1°C (Fig. 2A to D). BZP 323

significantly lowered body temperature in animals at 1 mg/kg dose, before handling (T0) revealing 324

hypothermia (H(3)=11.343, p=0.010, then p<0.05, Dunn’s test) (Fig. 2A). BZP dose-dependently 325

prevented stress-induced hyperthermia (F3,41=18.290, p<0.001) (Fig. 2B). Compared to the vehicle-326

treated animals, BZP was effective at doses of 0.5 and 1 mg/kg (p<0.05, Dunnett’s test). EFX also 327

induced changes in body temperature but without hypothermia at the highest dose (50 mg/kg) 328

compared with control animals (F3,41 =2.269, p=0.095) (Fig. 2C). As observed for BZP, the 329

temperature increase was dose-dependently prevented by EFX (H(3)=20.072, p<0.001) with a 330

significant effect at 25 and 50 mg/kg dose (p<0.05, Dunn’s test) (Fig. 2D). The anxiolytic effects of 331

BZP and EFX were also assessed by evaluating the behavioral approach in the presence of an 332

unfamiliar object. The duration of contact with a novel object was significantly decreased in animals 333

treated with BZP (H(3)=15.304, p=0.002) from the dose of 1 mg/kg (p<0.05, Dunn’s test) (Fig. 2E). 334

The same was observed with EFX (H(3)=17.536, p<0.001), with a significant effect observed at doses 335

of 25 and 50 mg/kg compared with the control animals (p<0.05, Dunn’s test) (Fig. 2F). In conclusion, 336

these two different behavioral tests led to similar anxiolytic doses of EFX (25-50 mg/kg, IP). 337

338

339

340

Page16of37

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

Fig. 2. Comparison of the effects of EFX and BZP on anxiety-related behaviors in mice. (A-D) The graphs show the 362 evolution of the mean rectal temperature (± SEM) at T0 and T0+10 min after treatment with vehicle (dose 0) or BZP (A) 363 or EFX (C) through IP route, 60 min before the first temperature measurement at T0. Histograms represent the mean (± 364 SEM) of the difference of the rectal temperature measured at T0 and T+10 min in the same mouse after treatment with 365 vehicle or BZP (B) or EFX (D) at the indicated doses. *p<0.05 compared with the vehicle group (Dunn test). (E,F) 366 Histograms illustrate the mean time (± SEM) of contact with an unfamiliar object after IP administration of BZP (E) or 367 EFX (F). *p<0.05 compared with the respective vehicle groups (dose 0). (see “data analysis and statistics” in the Material 368 and Methods section). Animal numbers are indicated inside the bars. 369

0 .0 0 .2 0 .4 0 .6 0 .8 1 .0 1 .23 5 .0

3 5 .5

3 6 .0

3 6 .5

3 7 .0

3 7 .5

3 8 .0

3 8 .5

3 9 .0

*

T 0

T + 1 0 m in

*

*

B Z P (m g /k g )

Re

cta

l te

mp

era

ture

(°C

)

0 1 0 2 0 3 0 4 0 5 0 6 03 6 .0

3 6 .5

3 7 .0

3 7 .5

3 8 .0

3 8 .5

3 9 .0T 0

T + 1 0 m in

*

E F X (m g /k g )

Re

cta

l te

mp

era

ture

(°C

)

0 0 .2 5 0 .5 1-1 .0

-0 .8

-0 .6

-0 .4

-0 .2

0 .0

0 .2

0 .4

0 .6

0 .8

1 .0

B Z P (m g /k g )

*

15 10 10 10

*

Diff

ere

nc

e i

n t

em

pe

ratu

re (

°C)

0 1 2 .5 2 5 5 00 .0

0 .1

0 .2

0 .3

0 .4

0 .5

0 .6

0 .7

0 .8

0 .9

1 .0

E F X (m g /k g )

*15 10 10 10

*

Diff

ere

nc

e i

n t

em

pe

ratu

re (

°C)

0 0 .2 5 0 .5 10

5 0

1 0 0

1 5 0

2 0 0

B Z P (m g /k g )

*12 10 10 10

Tim

e o

f co

nta

ct w

ith

th

e o

bje

ct (

s)

0 1 2 .5 2 5 5 00

5 0

1 0 0

1 5 0

2 0 0

2 5 0

E F X (m g /k g )

*

14 12 12 12

*

Tim

e o

f co

nta

ct w

ith

th

e o

bje

ct (

s)

A B

C D

E F

Page17of37

3.2 Motor performance assessment 370

We then compared the impact of EFX and BZP on locomotion performance and motor 371

coordination (Fig. 3). As illustrated in Fig. 3, BZP (IP route) decreased spontaneous locomotor 372

activity (H(3)=8.229, p=0.042). These effects were statistically significant for the 1 mg/kg dose 373

(p<0.05; Dunn’s test) (Fig. 3A). EFX (IP route) decreased spontaneous locomotor activity with a 374

significant difference at 100 mg/kg (H(4)=9.633, p=0.047) compared to control animals (p<0.05, 375

Dunn’s test) (Fig. 3B). BZP reduced the time on the rotarod (H(3)=9.167, p=0.027), with a significant 376

effect at the 1 mg/kg dose (p<0.05, Dunn’s test) (Fig. 3C). EFX was devoid of any effect up to the 377

100 mg/kg dose and affected motor coordination at the 150 mg/kg dose compared with control 378

animals (H(5)=19.006, p=0.002 then p<0.05, Dunn’s test) (Fig. 3D). In conclusion, BZP triggers 379

motor impairments at anxiolytic doses (1 mg/kg, IP), while EFX exhibits no locomotor effects at 380

efficient anxiolytic doses (25 to 50 mg/kg, IP). 381

Page18of37

382 Fig. 3. Comparison of BZP and EFX on locomotor activity and motor coordination. (A, B) Spontaneous locomotor 383 activities were assayed in the actimeter test in mice after IP injections of BZP (A) or EFX (B). The bars represent the 384 means ± SEM of the number of infrared beam interruptions over15 min. (C,D) Histograms illustrate the time on the rod 385 (mean ± SEM) after injection of BZP (C) or EFX (D). Animal numbers are indicated inside the bars. *p <0.05 compared 386 with the respective vehicle groups (see “data analysis and statistics” in the Material and Methods section). 387

3.3 EFX effects on GABA currents depends on α subunit isoforms 388

Because the distribution of GABAAR α subunits within the CNS is heterogeneous and 389

contributes to their receptor functions, we next investigated the involvement of α subunits in the EFX 390

mode of action using electrophysiology. To achieve this goal, we compared the effects of EFX (2 to 391

100 µM) on GABA-induced currents elicited by α1GABAARs, α2GABAARs, α3GABAARs, 392

α4GABAARs, α5GABAARs or α6GABAARs, containing β3 together with γ2S or d, when appropriate. 393

We first challenged EFX effects on GABA-currents elicited by synaptic α1β3γ2S, α2β3γ2S 394

and α3β3γ2S GABAARs expressed in Xenopus oocytes (Fig. 4). We observed that EFX displays both 395

0 0 .2 5 0 .5 10

5 0 0

1 0 0 0

1 5 0 0

2 0 0 0

2 5 0 0

3 0 0 0

3 5 0 0

4 0 0 0

B Z P (m g /k g )

*

1 0 1 0 1 0 1 0

Lo

co

mo

tor

act

ivity

(co

un

ts /

15

min

)

0 1 2 .5 2 5 5 0 1 0 00

5 0 0

1 0 0 0

1 5 0 0

2 0 0 0

2 5 0 0

3 0 0 0

E F X (m g /k g )

13 10 10 10

*

10

Lo

co

mo

tor

act

ivity

(co

un

ts /

15

min

)

0 0 .2 5 0 .5 10

2 0

4 0

6 0

8 0

1 0 0

1 2 0

B Z P (m g /k g )

*

1 0 1 0 1 0Tim

e o

n r

od

(s

)

12

0 1 2 .5 2 5 5 0 1 0 0 1 5 00

2 0

4 0

6 0

8 0

1 0 0

1 2 0

E F X (m g /k g )

13 10 10 10

*

10 10Tim

e o

n r

od

(s

)

A B

C D

Page19of37

agonist and potentiating effects as previously described [24]. The agonist effects of EFX were 396

revealed by its perfusion before co-application with GABA at EC10 (Table S2). EFX exhibits almost 397

no agonist effects on α1β3γ2S GABAARs (12.4 ± 5.8 and 23.7 ± 10.4 % of GABA EC10 at 50 and 100 398

µM EFX, respectively) (Fig. 4A). In contrast, comparable agonist effects were observed with 399

α2β3γ2S GABAARs (171 ± 24.9 % of GABA EC10, at 100 µM EFX, Fig. 4B) and α3β3γ2S GABAARs 400

(153.6 ± 27.3% of GABA EC10, at 100 µM EFX, Fig. 4C). In comparison with GABA, EFX exerted 401

weaker agonist effects (~100 fold less efficient). For these three receptors, EFX potentiation of GABA 402

EC10-evoked currents was dose-dependent, reaching a plateau at 50 µM (Fig. 4). At this concentration, 403

EFX induced a potentiation of GABA EC10-evoked currents by 68.9 ± 11.6%, 160.3 ± 40.2% and 404

410.7 ± 20.2% of α1β3γ2S, α2β3γ2S and α3β3γ2S, respectively. The potentiating effects induced 405

by EFX from 2 to 100 µM, was ~2.4-6.0-fold stronger (p<0.05) with α3β3γ2S than with α1β3γ2S 406

and α2β3γ2S GABAARs. Taking in account both agonist and potentiating effects, α1β3γ2S GABAARs 407

are found much less sensitive to EFX than α2β3γ2S and α3β3γ2S GABAARs. 408

Page20of37

Fig. 4. Effects of EFX on GABA-activated currents mediated by three synaptic GABAARs (a1β3γ2S, a2β3γ2S and 409 a3β3γ2S). EFX effects were investigated by TEVC in Xenopus oocytes expressing a1β3γ2S (A), a2β3γ2S (B) and 410 a3β3γ2S GABAARs (C). (A-C) Increasing concentrations of EFX (2, 20, 50 and 100 µM) were applied 2 min before co-411 application of GABA at EC10 (inbox). The amplitudes of EFX-evoked currents (● ) were normalized to the amplitude of 412 control currents (n) obtained with GABA alone at EC10. The potentiation effects of EFX was determined as the percentage 413 increase of EC10-GABA current amplitudes (● ). Left panel, GABA EC10-induced representative currents are illustrated, 414 showing the partial agonist and positive modulatory effects of EFX. Right panel, data points (mean ± SEM of 6-11 oocytes 415 from at least two different animals) were fitted by non-linear regression to the Hill equation with variable slope using 416 GraphPad Prism 7. Statistical analyses were performed using one-way ANOVA tests followed by Tukey’s post-hoc 417 correction (comparison with data obtained at 2 µM, *p<0.05, **p <0.01; ***p <0.001; ****p <0.0001, ns: not significant). 418

0

1 0 0

2 0 0

3 0 0

4 0 0

5 0 0

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

[E F X ] (µ M )

Po

ten

tia

tion

(%

, )

EF

X-e

voke

d cu

rren

ts (%, )

2 02 5 0 1 0 0

ns ns * ***ns

ns ns**

Aa1b3γ2S

*

control 20 µM 50 µM2 µM 100 µM

●

¨●// // // //

200 nA

2 min

n

200 nA

2 min

2 µMcontrol 20 µM 50 µM 100 µM

●

●// // // //n

a3b3γ2SC

B

0

1 0 0

2 0 0

3 0 0

4 0 0

5 0 0

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

[E F X ] (µ M )

Po

ten

tia

tio

n

(%,

)

EF

X-e

voke

d cu

rren

ts (%, )

ns

2 2 0 5 0 1 0 0

ns

**

**** ***

***** ****

0

1 0 0

2 0 0

3 0 0

4 0 0

5 0 0

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

[E F X ] (µ M )

Po

ten

tia

tio

n

(%,

)

EF

X-e

voke

d cu

rren

ts (%, )

2 02 5 0 1 0 0

ns***

**** ***

ns*

*** ****

control 20 µM 50 µM 100 µM2 µM

n

// // // //

a2b3γ2S

●

●

200 nA

2 min

2 min10-15 minwash

10-15 minwash

Page21of37

Based on pharmacokinetic data [30], we estimated that 20 µM matches the concentration of 419

free EFX in the mouse brain after injection of anxiolytic doses (25-50 mg/kg, Fig. 2). Thus, we 420

compared agonist and potentiating effects of 20 µM EFX on α(1-3) β3γ2S GABAARs (Fig. 5A and 421

B). EFX (20 µM) acts as a partial agonist on α2GABAARs and α3GABAARs, while these agonist 422

effects are not significant on α1GABAARs (Fig. 5A). As for the potentiating effects, α3GABAARs 423

were much more sensitive to EFX (245.6 ± 41.8%) than α1 (50.1 ± 12.5%) and α2 (101.7 ± 12.2% 424

increase) -containing GABAARs (Fig. 5B). In the synaptic cleft the GABA concentration rapidly rises 425

up to the millimolar range [48], we thus compared the GABA concentration-response relationships 426

in the presence and absence of 20 µM EFX at α1GABAARs, α2GABAARs and α3GABAARs (Fig. 5C, 427

Table 1). EFX induced a decrease of GABA EC50 with α1GABAARs and α2GABAARs in a similar 428

extent (~4-fold). In contrast, the GABA potency on α3GABAARs was increased by 20.6 (Fig. 5C). 429

Taken together, our electrophysiological data show that, EFX behaves as a selective PAM of 430

α3GABAARs at concentration equivalent to anxiolytic doses. 431

432

Page22of37

Fig. 5. Pharmacological profile of EFX over a(1-3)b3g2S GABAARs. Comparison of EFX effects at 20 µM corresponding 433 to anxiolytic doses. Detailed analysis of agonist (A) and potentiating (B) effects of 20 µM EFX on a1b3g2S, a2b3g2S 434 and a3b3g2S GABAARs. One-way ANOVA followed by Tukey post-hoc test was used for the analysis ((*p<0.05, **p 435 <0.01; ***p <0.001; ****p <0.0001). The number of recorded oocytes is indicated above or inside the bars. (C) 436 Concentration-response curves of GABA-evoked currents in the absence (open circles) and presence (close circles) of 20 437 µM EFX. Statistical analyses were performed using a non-parametric Mann & Whitney unpaired t-test. All data are 438 expressed as the mean ± SEM (n ≥ 6). 439

Table 1 440 Parameters of the GABA concentration-response relationship at a1b3g2S, a2b3g2S and a3b3g2S 441 GABAARs modulated by 20 µM EFX. 442

control + 20 µM EFX

GABAAR subtype EC50 (µM) nH EC50 (µM) nH

a1b3g2S 9.99 ± 0.96 1.06 ± 0.09 2.37 ± 0.11 1.41 ± 0.08

a2b3g2S 1.31 ± 0.12 1.69 ± 0.28 0.32 ± 0.03 1.14 ± 0.01

a3b3g2S 10.89 ± 0.61 1.34 ± 0.09 0.53 ± 0.08 1.05 ± 0.15

The concentration-response relationships were analyzed using the Hill-Langmuir equation with variable slope. nH : Hill 443 slope. The data are mean ± SEM of at least two independent experiments. 444

-9 -8 -7 -6 -5 -4 -30

2 0

4 0

6 0

8 0

1 0 0

L o g ( [G A B A ]) (M )

Cu

rre

nt

de

ns

ity

(% o

f m

ax.

)

-9 -8 -7 -6 -5 -4 -30

2 0

4 0

6 0

8 0

1 0 0

L o g ( [G A B A ]) (M )

Cu

rre

nt

de

ns

ity

(% o

f m

ax.

)

-9 -8 -7 -6 -5 -4 -30

2 0

4 0

6 0

8 0

1 0 0

L o g ( [G A B A ]) (M )

Cu

rre

nt

de

ns

ity

(% o

f m

ax.

)

Cur

rent

dens

ity(%

) EC50 fold change = 4.2p<0.001

a1b3γ2S a2b3γ2S a3b3γ2S

EC50 fold change = 4.1p<0.05

EC50 fold change = 20.6p<0.001

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

3 0 0

a 1 a 2 a 3

Po

ten

tiatio

n (

20

µM

EF

X,

%)

9 9 8

ns

******

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

3 0 0

a 1 a 2 a 3

Ag

on

ist

eff

ect

(2

0 µ

M E

FX

, %

)

410 76

*

ns**

BA

C

Page23of37

Next, we tested EFX on synaptic α4β3γ2S GABAARs and extrasynaptic α4β3d GABAARs. In 445

both cases, EFX did not exhibit any agonist effects nor significantly potentiate GABA-induced 446

currents, even at high concentrations (Fig. 6A,B). Conversely, DZP at 2 µM enhanced GABA currents 447

elicited by α4β3γ2S GABAARs, as previously reported [49], but it did not affect α4β3d GABAARs 448

(Fig. 6A,B). The extrasynaptic α5β3γ2S GABAARs appeared to be also insensitive to EFX at low 449

concentrations (2 and 20 µM), and weakly sensitive to EFX at 50 µM (54.26 ± 29.19%) (Fig. 6C and 450

D). At 100 µM, EFX effects were significantly increased but did not reach a plateau (212.0 ± 31.97%). 451

The α6 subunit was expressed with γ or d in accordance with the native GABAAR composition in the 452

cerebellum [50]. Similar agonist effects were observed with α6β3γ2S GABAARs (76.9 ± 20.3% of 453

GABA EC10, at 100 µM EFX,) and α6β3δ GABAARs (46.8 % of GABA EC10, at 100 µM EFX) (p=0.23, 454

Mann and Whitney test) (Fig. 6E,F) as seen with α2GABAARs and α3GABAARs (Fig. 4B,C). 455

Moreover, the PAM effects of EFX revealed equal sensitivities of synaptic α6β3γ2SGABAARs and 456

extrasynaptic α6β3δ GABAARs. EFX-potentiation of GABA-evoked currents reached a plateau at 50 457

µM EFX (170.5 ± 48.4% increase for α6β3γ2S GABAARs and 143.6 ± 22.3% for α6β3δ GABAARs, 458

p>0.99, Mann and Whitney test) in accordance with a specific concentration-dependent mode of 459

action (Fig. 6E and F). 460

461

Page24of37

Fig. 6. Effects of EFX on GABA EC10-activated currents mediated by synaptic (a4β3γ2S and a6 β3γ2S) and extrasynaptic 462 (a4β3δ, a5β3γ2S and a6β3δ) GABAARs. (A) Superimposed current traces evoked by GABA EC10 in a representative cell 463 expressing a4β3γ2S or a4β3δ GABAARs in the absence (black traces) or presence of DZP (2 µM) or EFX. (B) Graphs 464 illustrating the mean (± SEM) EFX potentiation of GABA EC10-activated currents mediated by synaptic and extrasynaptic 465 a4GABAARs. (C) Current traces evoked by GABA EC10 in a representative oocyte expressing extrasynaptic a5β3γ2S 466 GABAARs in the absence (black trace) or presence of EFX (2 to 100 µM). (D) Graphs showing the mean (± SEM) EFX 467 potentiation of GABA EC10-evoked currents. Statistical analyses were performed by one-way ANOVA followed by Tukey 468 post-hoc test (*p<0.05, **p <0.01; ***p <0.001; ****p <0.0001, ns: not significant). (E) Current traces evoked by GABA 469 EC10 with synaptic a6β3γ2S and extrasynaptic a6β3δ GABAARs in the absence (black traces) or presence of increasing 470 concentrations of EFX (2 to 100 µM). (F) Graphs showing the mean (± SEM) EFX potentiation of GABA EC10-activated 471 currents mediated by a6β3γ2S and a6β3δ GABAARs. Statistical analyses were performed using a non-parametric Mann 472 & Whitney unpaired t-test (* p<0.05; ns: not significant). 473

0

1 0 0

2 0 0

3 0 0

0

1 0 0

2 0 0

3 0 0

EF

X-e

voke

d cu

rren

ts (%, _

)

[E F X ] (µ M )

Po

ten

tiatio

n (

%,_

)

2 2 0 5 0 1 0 0

a 6 b 3 d

a 6 b 3 g 2 s

*

*** ***

ns ***** **

ns

******* ****

ns * ***

[E F X ] (µ M )

Po

ten

tia

tio

n (

%,

)

EF

X-e

voke

d cu

rren

ts (%, )

0

1 0 0

2 0 0

3 0 0

0

1 0 0

2 0 0

3 0 0

2 2 0 5 0 1 0 0

ns nsns

*

ns ns ** ****

0

1 0 0

2 0 0

3 0 0

[E F X ] (µ M )

Po

ten

tia

tio

n (

%)

a 4 b 3 g 2

a 4 b 3 d

2 2 0 5 0 1 0 0

ns ns ns nsns ns

ns ns

B

FE

A

a5b3g2S

100 µM

20 µM2 µM

50 µM250 nA

1 min

DC

EFX 100 µM

DZP 2 µM

a4b3d

EFX 2 µM

DZP 2 µMEFX 100 µM

a4b3γ2S

30 s20 nA

EFX 2 µM

a6b3da6b3γ2S

250 nA1 min

2 µM

20 µM

50 µM100 µM

EFX

EFX

Page25of37

In conclusion, taking into account both agonist and potentiating EFX effects, GABAARs can 474

be ranked in three categories: i) resistant (α4GABAARs and α5GABAARs), ii) moderately sensitive 475

(α1GABAARs, α2GABAARs and α6GABAARs) and iii) highly sensitive (α3GABAARs) to EFX. 476

3.7 Modelling of EFX-α3β3γ2s GABAAR interaction 477

To gain further insight into the mechanism of action of EFX, we generated a homology model 478

of α3β3γ2s GABAAR (Fig. 7A) to predict how EFX binds to its receptor site. The resulting computed 479

docking model was consistent with an EFX binding site located between α and β subunits in the 480

extracellular domain (Fig. 7B and C). The pocket found at the interface between α3 and β3 subunits, 481

homologous to the GABA binding sites, was large enough to accommodate EFX. Among the putative 482

binding modes, one was found in which EFX bound in the proximity of five amino acid residues of 483

the β3 subunit: N66, Y87, Q89, Y91 and R194 (Fig. 7D). In α3 subunit, we identified two amino acid 484

residues (S257, S258) that may be involved in the EFX-GABAAR interaction. Two residues of β3, 485

N66 and R194, are conserved in β2 and β3 and are different in β1 subunits (R66 and N194). This pair 486

of amino acids might therefore control the binding mode of EFX as variations at these residues might 487

explain the subunit selectivity. 488

489

Page26of37

Fig. 7. Binding modes of EFX obtained by docking on the mouse α3β3γ2S GABAAR. (A) Model of the receptor viewed 490 from the membrane plane. The protein is shown in cartoon representation with a different color code for each polypeptide. 491 The position of the membrane is represented by a sphere positioned at the level of lipid head groups as determined by the 492 Orientations of Proteins in Membranes database [51]. (B-D) Binding mode of EFX by docking on the mouse 493 α3β3γ2SGABAAR model. EFX (CPK representation) interacts with a pocket localized at the α3 (cyan) β3 (magenta) 494 interface, homologous to GABA binding sites. The binding sites of BZD, GABA and EFX are indicated by arrows (B). 495 Lateral view of the extracellular domain (C). Close-up showing the EFX-binding pocket (EFX appears in sticks) (D). 496

A

C

EFX

α3

γ2s

β3

β3α3

B

90°

D

45°

β3α3

BZD

GABAGABA

Page27of37

4. Discussion 497

The current study shows that a single administration EFX (25-50 mg/kg) induced a robust 498

anxiolytic behavior in mice subjected to stress-induced hyperthermia and novel object exploration 499

tests. In the same range of doses, and unlike classical BZDs, EFX did not evoke any secondary effects 500

in the spontaneous locomotor activity and rotarod performance. Pharmacokinetic data in Balb/cByJ 501

mice treated with anxiolytic doses indicated that EFX brain content reaches concentration range from 502

16 to 31 µM [30]. In addition, EFX exhibits lipophilic properties with an estimated partition 503

coefficient (log P) of 2 and a brain/ plasma ratio range of 2.2-2.9 [52]. Based on different reports in 504

the literature, it is reasonable to assume that an equilibrium between the total and free fractions which 505

depends on physicochemical properties of the compound occurs in the brain tissue [53-55] and, as a 506

result this could support the relevance of the effective concentrations in the present 507

electrophysiological studies. In this context, we demonstrated that the a subunit plays an important 508

role in EFX-induced positive effects on GABAARs. EFX favours GABA potency over GABAARs with 509

the following rank order: α3b3g2S > α2b3g2S > α6b3g2S and no or weak effects on α1b3g2S, 510

α4b3g2S and α5b3g2S. 511

4.1 EFX displays anxiolytic properties with weak side effects 512

Our findings confirm the anxiolytic properties of EFX at similar doses to those previously 513

used in other anxiety mouse models [30]. In comparison, EFX displays approximately 50-fold less 514

potent anxiolytic effects than BZP and DZP [8,9]. However, both BZP and DZP strongly alter 515

locomotor performance and awakening at anxiolytic doses, while EFX does not. It is noteworthy that 516

pharmacokinetic factors involving, for example, active metabolites or differences in the extent of 517

metabolism could explain the differences in the effective doses of EFX and BZP. We reasoned that 518

BZDs, which have a high potency (submicromolar) for enhancing GABA-evoked currents, will 519

produce effects at a lower concentration than EFX which has a lower potency (micromolar) for 520

Page28of37

GABAARs. Interestingly, EFX exhibits higher efficacy for α2GABAARs and α3GABAARs, known to 521

mediate anxiolytic effects (up to 171% and 410% for α2GABAARs and α3GABAARs, respectively) 522

than DZP (108% and 160% for α2GABAARs and α3GABAARs, respectively) [56]. 523

Compelling evidence indicates that the anxiolytic effects of BZP cannot be dissociated from 524

its sedative and myorelaxant effects, while the therapeutic margin is wider with EFX. In addition, 525

previous results have shown that EFX is devoid of amnesic effects at anxiolytic doses (50 mg/kg, IP 526

route) in the rat [57]. On the other hand, BZP and DZP display amnesic activity at doses producing 527

anxiolytic effects (from 0.25mg/kg, IP) in the mouse [58]. As observed in rodents, patients treated 528

with EFX for adjustment disorders with anxiety do not exhibit adverse effects, such as the memory 529

and vigilance disturbances [19,59,60]. This is perfectly in line with our electrophysiological data 530

showing the absence of effects of EFX on a4GABAARs and a5GABAARs, known to be involved in 531

cognitive functions [4]. 532

We cannot rule out the possibility that EFX anxiolytic properties rely on both direct and 533

indirect GABAAR stimulation mechanisms. Since EFX has been shown to stimulate the synthesis of 534

neurosteroids, such as allopregnenolone, which directly boosts GABAAR activity [27,43,61], this may 535

account for its anxiolytic effects. Neurosteroids equally enhance GABA-evoked currents mediated 536

by a1GABAARs, a3GABAARs and a6GABAARs, while they have little effects on α2GABAARs, 537

α4GABAARs and α5GABAARs [58,62], suggesting that EFX should induce both sedation and 538

anxiolysis. However, because EFX did not induce sedation at anxiolytic doses, this minors the 539

involvement of neurosteroids in the EFX mode of action. We hypothesize that EFX may exert its 540

anxiolytic effect through a direct enhancement of the activity of GABAARs. To date, there is no 541

experimental data on EFX modulation of GABAARs to conclude a plausible mode of action. However, 542

using recombinant murine GABAARs expressed in Xenopus oocytes, it has been shown that both 543

efficacy and potency of GABA are enhanced by EFX [24]. The effect of EFX might be explained by 544

Page29of37

either mechanism, i.e. an increased frequency of the open state of GABAARs and/or an increase of the 545

duration of burst openings. 546

Further studies using a chronic treatment are warranted to support the specificity of EFX 547

compared to BZDs in the development of tolerance complex phenomenom involving in part selective 548

alterations in GABAAR receptor subunit expression [63]. 549

4.2 The EFX mode of action depends on the GABAAR α subunits 550

Our electrophysiological data demonstrate that EFX behaves both as a partial agonist and a 551

PAM of GABAARs. In fact, EFX strongly potentiates GABA-evoked currents mediated by 552

α2GABAARs, α3GABAARs and/or α6GABAARs, with major effects on α3GABAARs in comparison to 553

any other GABAARs. This was also highlighted by a larger enhancement of the GABA potency on 554

α3GABAARs, than on α1GABAARs and α2GABAARs. However, the involvement of α3GABAARs in 555

anxiolysis is still a matter of debate. As mentioned above, there is still a controversy concerning the 556

implication of α2GABAARs, α3GABAARs and α5GABAARs in the control of anxiety-related 557

behaviors [14,15,17,64]. Other non-BZD compounds such as TPA023, AZD6280 and AZD7325, 558

have been shown to exert anxiolysis without sedative side effects in rodents and/or humans by 559

preferentially enhancing α2GABAARs and a3GABAARs over the other GABAAR subtypes [64,65]. 560

However, these three compounds bind to the BZD site, while EFX does not [23]. In addition, another 561

non-BZD compound, TP003 was first reported as a selective PAM of α3GABAAR, and initially 562

considered to exhibit anxiolytic properties through this receptor [14,66]. However, two recent studies 563

have revealed that this drug is not selective to α3GABAAR, but equally modulates GABA-evoked 564

currents mediated by α5GABAAR and also moderately potentiates α1GABAARs and α2GABAAR 565

[65,67]. TP003 was indeed shown to counteract anxiety behaviors in both rodents and squirrel 566

monkeys and thus highlights the medical use of α3GABAAR-selective molecules as efficient 567

Page30of37

anxiolytics with no sedative secondary effects [14,66]. Therefore, we believe that the anxiolytic-like 568

effects of EFX in mice are related to the modulation of α2GABAARs and α3GABAARs. 569

4.3 The EFX binding site 570

Our objective was to challenge the possible influence of the α subunit for at least three reasons. 571

First, β2 homopentamers are less sensitive to EFX than β3 homopentamers. However, when they are 572

combined with α1 or α2 subunits, the resulting binary GABAARs display a different pharmacological 573

profile: α1β2-3 and α2β2-3 GABAARs are equally potentiated, indicating that α1 and α2 subunits 574

modulate EFX potentiation [24]. Second, α1 and α2 subunits share a high amino acid sequence 575

identity, while α4-6 are structurally more distant [68] and thus could potentially have distinct 576

pharmacological influences. Here, we bring compelling evidences demonstrating that stimulating 577

effects of EFX are much stronger on α3β3γ2S than on α2β3γ2S, α6β3γ2S and α1β3γ2S GABAARs, 578

while α4β3γ2S and α5β3γ2S are almost insensitive. Altogether, our findings indicate a strong 579

regulatory effect of the α subunit on EFX mode of action. 580

We also examined the involvement of γ2S and d subunits in the mode of action of EFX and 581

we observed that α4β3γ2S and α4β3d on one hand, α6β3γ2S and α6β3d GABAARs on the other hand, 582

are equally sensitive to EFX. This reinforces the idea that, unlike BZDs [7], the third subunit is not 583

involved in the EFX-GABAAR interaction. 584

Consecutively, we hypothesized that EFX site is likely located between the α and β subunits. 585

Our 3D docking simulation suggests that EFX binds in a pocket at the α/β subunit interface 586

homologous to the GABA binding pocket recently described [69], highlighting putative amino acid 587

residues involved in EFX binding. Interestingly, among them, two residues of β3, N66 and R194, are 588

conserved in β2 and β3 and differ in the β1 subunits (R66 and N194). This pair of amino acids may 589

belong to the binding site of EFX and summarise its subunit selectivity. Site-directed mutagenesis 590

experiments are required to validate this hypothesis and to define the residues underlying the EFX 591

Page31of37

selectivity towards α3GABAARs. These experiments will allow us to construct genetic models in 592

which specific α(1-6)GABAARs subtypes will be rendered insensitive to EFX to directly correlate 593

specific α2GABAARs or α3GABAARs to its anxiolytic effects or test whether α3GABAAR functions 594

are involved in the regulation of anxiety. 595

5. Conclusions 596

In conclusion, this study provides new information about the mode of action of EFX, a non-597

BZD anxiolytic compound, showing that it potentiates GABA transmission, mainly through the 598

interaction with α2GABAARs and α3GABAARs and likely their associated functions. Modelling 599

simulation indicates that EFX could interact with a pocket localized at the α/β subunits 600

interface,homologous to the GABA binding site. To the best of our knowledge, EFX belongs to the 601

group of non-BZD molecules which act at a site distinct from the classical BZDs site and exert 602

positive effects on anxiety without secondary effects. EFX may therefore serve as a molecular 603

template for the design of novel anxiolytics with similar mechanisms of action and higher potency. 604

605

Conflict of interest 606

The authors declare no conflicts of interest. 607

608

Acknowledgements 609

We are indebted to Dr. Alain Hamon and Bastien Faucard for helpful discussions about the work. We 610

are grateful to Sophie Quinchard for technical assistance. We also thank Dr. Ulrich Jarry, Charlène 611

Labège, Céline Alamichel and Vincent Juif for their contribution to preparing the cDNAs encoding 612

α3-6 subunits in the pRK5 vector and electrophysiology experiments. 613

614

Page32of37

References 615

[1] R.W. Olsen, W. Sieghart, International Union of Pharmacology. LXX. Subtypes of gamma-616

aminobutyric acid(A) receptors: classification on the basis of subunit composition, pharmacology, 617

and function. Update, Pharmacol Rev 60(3) (2008) 243-60. 618

[2] M. Farrant, Z. Nusser, Variations on an inhibitory theme: phasic and tonic activation of 619

GABA(A) receptors, Nat Rev Neurosci 6(3) (2005) 215-29. 620

[3] R.W. Olsen, W. Sieghart, GABA A receptors: subtypes provide diversity of function and 621

pharmacology, Neuropharmacology 56(1) (2009) 141-8. 622

[4] E. Engin, R.S. Benham, U. Rudolph, An Emerging Circuit Pharmacology of GABAA Receptors, 623

Trends Pharmacol Sci 39(8) (2018) 710-732. 624

[5] T. Atkin, S. Comai, G. Gobbi, Drugs for Insomnia beyond Benzodiazepines: Pharmacology, 625

Clinical Applications, and Discovery, Pharmacol Rev 70(2) (2018) 197-245. 626

[6] M.J. Brodie, F. Besag, A.B. Ettinger, M. Mula, G. Gobbi, S. Comai, A.P. Aldenkamp, B.J. 627

Steinhoff, Epilepsy, Antiepileptic Drugs, and Aggression: An Evidence-Based Review, Pharmacol 628

Rev 68(3) (2016) 563-602. 629

[7] D.B. Pritchett, H. Sontheimer, B.D. Shivers, S. Ymer, H. Kettenmann, P.R. Schofield, P.H. 630

Seeburg, Importance of a novel GABAA receptor subunit for benzodiazepine pharmacology, Nature 631

338(6216) (1989) 582-5. 632

[8] L. Steru, B. Thierry, R. Chermat, B. Millet, P. Simon, R.D. Porsolt, Comparing benzodiazepines 633

using the staircase test in mice, Psychopharmacology (Berl) 92(1) (1987) 106-9. 634

[9] M. Bourin, M. Hascoet, B. Mansouri, M.C. Colombel, J. Bradwejn, Comparison of behavioral 635

effects after single and repeated administrations of four benzodiazepines in three mice behavioral 636

models, J Psychiatry Neurosci 17(2) (1992) 72-7. 637

[10] U. Rudolph, F. Knoflach, Beyond classical benzodiazepines: novel therapeutic potential of 638

GABAA receptor subtypes, Nat Rev Drug Discov 10(9) (2011) 685-97. 639

[11] U. Rudolph, F. Crestani, D. Benke, I. Brunig, J.A. Benson, J.M. Fritschy, J.R. Martin, H. 640

Bluethmann, H. Mohler, Benzodiazepine actions mediated by specific gamma-aminobutyric 641

acid(A) receptor subtypes, Nature 401(6755) (1999) 796-800. 642

[12] R.M. McKernan, T.W. Rosahl, D.S. Reynolds, C. Sur, K.A. Wafford, J.R. Atack, S. Farrar, J. 643

Myers, G. Cook, P. Ferris, L. Garrett, L. Bristow, G. Marshall, A. Macaulay, N. Brown, O. Howell, 644

K.W. Moore, R.W. Carling, L.J. Street, J.L. Castro, C.I. Ragan, G.R. Dawson, P.J. Whiting, 645

Sedative but not anxiolytic properties of benzodiazepines are mediated by the GABA(A) receptor 646

alpha1 subtype, Nat Neurosci 3(6) (2000) 587-92. 647

Page33of37

[13] K. Löwe, F. Crestani, R. Keist, D. Benke, I. Brunig, J.A. Benson, J.M. Fritschy, T. Rulicke, H. 648

Bluethmann, H. Mohler, U. Rudolph, Molecular and neuronal substrate for the selective attenuation 649

of anxiety, Science 290(5489) (2000) 131-4. 650

[14] R. Dias, W.F. Sheppard, R.L. Fradley, E.M. Garrett, J.L. Stanley, S.J. Tye, S. Goodacre, R.J. 651

Lincoln, S.M. Cook, R. Conley, D. Hallett, A.C. Humphries, S.A. Thompson, K.A. Wafford, L.J. 652

Street, J.L. Castro, P.J. Whiting, T.W. Rosahl, J.R. Atack, R.M. McKernan, G.R. Dawson, D.S. 653

Reynolds, Evidence for a significant role of alpha 3-containing GABAA receptors in mediating the 654

anxiolytic effects of benzodiazepines, J Neurosci 25(46) (2005) 10682-8. 655

[15] J.R. Atack, GABAA receptor alpha2/alpha3 subtype-selective modulators as potential 656

nonsedating anxiolytics, Curr Top Behav Neurosci 2 (2010) 331-60. 657

[16] M.M. Savic, S. Huang, R. Furtmuller, T. Clayton, S. Huck, D.I. Obradovic, N.D. Ugresic, W. 658

Sieghart, D.R. Bokonjic, J.M. Cook, Are GABAA receptors containing alpha5 subunits contributing 659

to the sedative properties of benzodiazepine site agonists?, Neuropsychopharmacology 33(2) (2008) 660

332-9. 661

[17] L.M. Behlke, R.A. Foster, J. Liu, D. Benke, R.S. Benham, A.J. Nathanson, B.K. Yee, H.U. 662

Zeilhofer, E. Engin, U. Rudolph, A Pharmacogenetic 'Restriction-of-Function' Approach Reveals 663

Evidence for Anxiolytic-Like Actions Mediated by alpha5-Containing GABAA Receptors in Mice, 664

Neuropsychopharmacology 41(10) (2016) 2492-501. 665

[18] J.R. Boissier, P. Simon, M. Zaczinska, J. Fichelle, [Experimental psychopharmacologic study 666

of a new psychotropic drug, 2-ethylamino-6-chloro-4-methyl-4-phenyl-4H-3,1-benzoxazine], 667

Therapie 27(2) (1972) 325-38. 668

[19] N. Nguyen, E. Fakra, V. Pradel, E. Jouve, C. Alquier, M.E. Le Guern, J. Micallef, O. Blin, 669

Efficacy of etifoxine compared to lorazepam monotherapy in the treatment of patients with 670

adjustment disorders with anxiety: a double-blind controlled study in general practice, Hum 671

Psychopharmacol 21(3) (2006) 139-49. 672

[20] D.J. Stein, Etifoxine versus alprazolam for the treatment of adjustment disorder with anxiety: a 673

randomized controlled trial, Adv Ther 32(1) (2015) 57-68. 674

[21] M. Aouad, N. Petit-Demouliere, Y. Goumon, P. Poisbeau, Etifoxine stimulates 675

allopregnanolone synthesis in the spinal cord to produce analgesia in experimental 676

mononeuropathy, Eur J Pain 18(2) (2014) 258-68. 677

[22] E. Simon-O'Brien, D. Gauthier, V. Riban, M. Verleye, Etifoxine improves sensorimotor deficits 678

and reduces glial activation, neuronal degeneration, and neuroinflammation in a rat model of 679

traumatic brain injury, J Neuroinflammation 13(1) (2016) 203. 680

Page34of37

[23] R. Schlichter, V. Rybalchenko, P. Poisbeau, M. Verleye, J. Gillardin, Modulation of 681

GABAergic synaptic transmission by the non-benzodiazepine anxiolytic etifoxine, 682

Neuropharmacology 39(9) (2000) 1523-35. 683

[24] A. Hamon, A. Morel, B. Hue, M. Verleye, J.M. Gillardin, The modulatory effects of the 684

anxiolytic etifoxine on GABA(A) receptors are mediated by the beta subunit, Neuropharmacology 685

45(3) (2003) 293-303. 686

[25] M. Verleye, Y. Akwa, P. Liere, N. Ladurelle, A. Pianos, B. Eychenne, M. Schumacher, J.M. 687

Gillardin, The anxiolytic etifoxine activates the peripheral benzodiazepine receptor and increases 688

the neurosteroid levels in rat brain, Pharmacol Biochem Behav 82(4) (2005) 712-20. 689

[26] B. Costa, C. Cavallini, E. Da Pozzo, S. Taliani, F. Da Settimo, C. Martini, The Anxiolytic 690

Etifoxine Binds to TSPO Ro5-4864 Binding Site with Long Residence Time Showing a High 691

Neurosteroidogenic Activity, ACS Chem Neurosci 8(7) (2017) 1448-1454. 692

[27] P. Liere, A. Pianos, J.P. Oudinet, M. Schumacher, Y. Akwa, Differential effects of the 18-kDa 693

translocator protein (TSPO) ligand etifoxine on steroidogenesis in rat brain, plasma and 694

steroidogenic glands: Pharmacodynamic studies, Psychoneuroendocrinology 83 (2017) 122-134. 695

[28] M. Verleye, R. Schlichter, J.M. Gillardin, Interactions of etifoxine with the chloride channel 696

coupled to the GABA(A) receptor complex, Neuroreport 10(15) (1999) 3207-10. 697

[29] M. Verleye, I. Heulard, J.M. Gillardin, Investigation of the anticonvulsive effect of acute 698

immobilization stress in anxious Balb/cByJ mice using GABA A-related mechanistic probes, 699

Psychopharmacology (Berl) 197(4) (2008) 523-34. 700

[30] M. Verleye, I. Heulard, P. Nuss, J.M. Gillardin, Effects of stress and etifoxine on pentobarbital-701

induced loss of righting reflex in Balb/cByJ and C57BL/6J mice, Neurosci Lett 353(2) (2003) 127-702

30. 703

[31] M. Sansone, Benzodiazepines and amphetamine on avoidance behaviour in mice, Arch Int 704

Pharmacodyn Ther 218(1) (1975) 125-32. 705

[32] R.B. Rastogi, Y.D. Lapierre, R.L. Singhal, Effect of a new benzodiazepine bromazepam on 706

locomotor performance and brain monoamine metabolism, J Neural Transm 42(4) (1978) 251-61. 707

[33] M. Verleye, S. Dumas, I. Heulard, N. Krafft, J.M. Gillardin, Differential effects of etifoxine on 708

anxiety-like behaviour and convulsions in BALB/cByJ and C57BL/6J mice: any relation to 709

overexpression of central GABAA receptor beta2 subunits?, Eur Neuropsychopharmacol 21(6) 710

(2011) 457-70. 711

[34] B. Olivier, T. Zethof, T. Pattij, M. van Boogaert, R. van Oorschot, C. Leahy, R. Oosting, A. 712

Bouwknecht, J. Veening, J. van der Gugten, L. Groenink, Stress-induced hyperthermia and anxiety: 713

pharmacological validation, Eur J Pharmacol 463(1-3) (2003) 117-32. 714

Page35of37

[35] T.J. Zethof, J.A. Van der Heyden, J.T. Tolboom, B. Olivier, Stress-induced hyperthermia in 715

mice: a methodological study, Physiol Behav 55(1) (1994) 109-15. 716

[36] J.A. Van der Heyden, T.J. Zethof, B. Olivier, Stress-induced hyperthermia in singly housed 717

mice, Physiol Behav 62(3) (1997) 463-70. 718

[37] S.F. De Boer, J.M. Koolhaas, Defensive burying in rodents: ethology, neurobiology and 719

psychopharmacology, Eur J Pharmacol 463(1-3) (2003) 145-61. 720

[38] A. Walker, J. Taylor, D. Rowe, D. Summers, A method for generating sticky-end PCR products 721

which facilitates unidirectional cloning and the one-step assembly of complex DNA constructs, 722

Plasmid 59(3) (2008) 155-62. 723

[39] S.P. Venkatachalan, J.D. Bushman, J.L. Mercado, F. Sancar, K.R. Christopherson, A.J. Boileau, 724

Optimized expression vector for ion channel studies in Xenopus oocytes and mammalian cells using 725

alfalfa mosaic virus, Pflugers Arch 454(1) (2007) 155-63. 726

[40] A. Draguhn, T.A. Verdorn, M. Ewert, P.H. Seeburg, B. Sakmann, Functional and molecular 727

distinction between recombinant rat GABAA receptor subtypes by Zn2+, Neuron 5(6) (1990) 781-8. 728

[41] C. Notredame, D.G. Higgins, J. Heringa, T-Coffee: A novel method for fast and accurate 729

multiple sequence alignment, J Mol Biol 302(1) (2000) 205-17. 730

[42] A. Sali, T.L. Blundell, Comparative protein modelling by satisfaction of spatial restraints, J 731

Mol Biol 234(3) (1993) 779-815. 732

[43] G.G. Krivov, M.V. Shapovalov, R.L. Dunbrack, Jr., Improved prediction of protein side-chain 733

conformations with SCWRL4, Proteins 77(4) (2009) 778-95. 734

[44] B.R. Brooks, C.L. Brooks, 3rd, A.D. Mackerell, Jr., L. Nilsson, R.J. Petrella, B. Roux, Y. Won, 735

G. Archontis, C. Bartels, S. Boresch, A. Caflisch, L. Caves, Q. Cui, A.R. Dinner, M. Feig, S. 736

Fischer, J. Gao, M. Hodoscek, W. Im, K. Kuczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci, 737

R.W. Pastor, C.B. Post, J.Z. Pu, M. Schaefer, B. Tidor, R.M. Venable, H.L. Woodcock, X. Wu, W. 738

Yang, D.M. York, M. Karplus, CHARMM: the biomolecular simulation program, J Comput Chem 739

30(10) (2009) 1545-614. 740

[45] J. Henin, R. Salari, S. Murlidaran, G. Brannigan, A predicted binding site for cholesterol on the 741

GABAA receptor, Biophys J 106(9) (2014) 1938-49. 742

[46] O. Trott, A.J. Olson, AutoDock Vina: improving the speed and accuracy of docking with a new 743

scoring function, efficient optimization, and multithreading, J Comput Chem 31(2) (2010) 455-61. 744

[47] R.R. Ugale, A.N. Sharma, D.M. Kokare, K. Hirani, N.K. Subhedar, C.T. Chopde, Neurosteroid 745

allopregnanolone mediates anxiolytic effect of etifoxine in rats, Brain Res 1184 (2007) 193-201. 746

Page36of37

[48] A. Barberis, E.M. Petrini, J.W. Mozrzymas, Impact of synaptic neurotransmitter concentration 747

time course on the kinetics and pharmacological modulation of inhibitory synaptic currents, Front 748

Cell Neurosci 5 (2011) 6. 749

[49] H. You, J.L. Kozuska, I.M. Paulsen, S.M. Dunn, Benzodiazepine modulation of the rat 750

GABAA receptor alpha4beta3gamma2L subtype expressed in Xenopus oocytes, 751

Neuropharmacology 59(6) (2010) 527-33. 752

[50] Z. Nusser, W. Sieghart, P. Somogyi, Segregation of different GABAA receptors to synaptic and 753

extrasynaptic membranes of cerebellar granule cells, J Neurosci 18(5) (1998) 1693-703. 754

[51] M.A. Lomize, I.D. Pogozheva, H. Joo, H.I. Mosberg, A.L. Lomize, OPM database and PPM 755

web server: resources for positioning of proteins in membranes, Nucleic Acids Res 40(Database 756

issue) (2012) D370-6. 757

[52] C. Bouillot, F. Bonnefoi, F. Liger, L. Zimmer, A microPET comparison of the effects of 758

etifoxine and diazepam on [(11)C]flumazenil uptake in rat brains, Neurosci Lett 612 (2016) 74-9. 759

[53] P.L. Toutain, A. Bousquet-Melou, Free drug fraction vs free drug concentration: a matter of 760

frequent confusion, J Vet Pharmacol Ther 25(6) (2002) 460-3. 761

[54] T.S. Maurer, D.B. Debartolo, D.A. Tess, D.O. Scott, Relationship between exposure and 762

nonspecific binding of thirty-three central nervous system drugs in mice, Drug Metab Dispos 33(1) 763

(2005) 175-81. 764

[55] S.G. Summerfield, K. Read, D.J. Begley, T. Obradovic, I.J. Hidalgo, S. Coggon, A.V. Lewis, 765

R.A. Porter, P. Jeffrey, Central nervous system drug disposition: the relationship between in situ 766

brain permeability and brain free fraction, J Pharmacol Exp Ther 322(1) (2007) 205-13. 767

[56] A. Lippa, P. Czobor, J. Stark, B. Beer, E. Kostakis, M. Gravielle, S. Bandyopadhyay, S.J. 768

Russek, T.T. Gibbs, D.H. Farb, P. Skolnick, Selective anxiolysis produced by ocinaplon, a 769

GABA(A) receptor modulator, Proc Natl Acad Sci U S A 102(20) (2005) 7380-5. 770

[57] M. Verleye, J.M. Gillardin, Effects of etifoxine on stress-induced hyperthermia, freezing 771

behavior and colonic motor activation in rats, Physiol Behav 82(5) (2004) 891-7. 772

[58] R.D. Porsolt, A. Lenegre, I. Avril, G. Doumont, Antagonism by exifone, a new cognitive 773

enhancing agent, of the amnesias induced by four benzodiazepines in mice, Psychopharmacology 774

(Berl) 95(3) (1988) 291-7. 775

[59] J. Micallef, C. Soubrouillard, F. Guet, M.E. Le Guern, C. Alquier, B. Bruguerolle, O. Blin, A 776

double blind parallel group placebo controlled comparison of sedative and mnesic effects of 777

etifoxine and lorazepam in healthy subjects [corrected], Fundam Clin Pharmacol 15(3) (2001) 209-778

16. 779

Page37of37

[60] D. Deplanque, F. Machuron, N. Waucquier, E. Jozefowicz, S. Duhem, S. Somers, O. Colin, A. 780

Duhamel, R. Bordet, Etifoxine impairs neither alertness nor cognitive functions of the elderly: A 781

randomized, double-blind, placebo-controlled crossover study, Eur Neuropsychopharmacol 28(8) 782

(2018) 925-932. 783

[61] L. Wolf, A. Bauer, D. Melchner, H. Hallof-Buestrich, P. Stoertebecker, E. Haen, M. Kreutz, N. 784

Sarubin, V.M. Milenkovic, C.H. Wetzel, R. Rupprecht, C. Nothdurfter, Enhancing neurosteroid 785

synthesis--relationship to the pharmacology of translocator protein (18 kDa) (TSPO) ligands and 786

benzodiazepines, Pharmacopsychiatry 48(2) (2015) 72-7. 787

[62] D. Belelli, A. Casula, A. Ling, J.J. Lambert, The influence of subunit composition on the 788

interaction of neurosteroids with GABA(A) receptors, Neuropharmacology 43(4) (2002) 651-61. 789

[63] M.C. Gravielle, Activation-induced regulation of GABAA receptors: Is there a link with the 790

molecular basis of benzodiazepine tolerance?, Pharmacol Res 109 (2016) 92-100. 791

[64] J.R. Atack, K.A. Wafford, S.J. Tye, S.M. Cook, B. Sohal, A. Pike, C. Sur, D. Melillo, L. 792

Bristow, F. Bromidge, I. Ragan, J. Kerby, L. Street, R. Carling, J.L. Castro, P. Whiting, G.R. 793

Dawson, R.M. McKernan, TPA023 [7-(1,1-dimethylethyl)-6-(2-ethyl-2H-1,2,4-triazol-3-794

ylmethoxy)-3-(2-fluorophenyl )-1,2,4-triazolo[4,3-b]pyridazine], an agonist selective for alpha2- 795

and alpha3-containing GABAA receptors, is a nonsedating anxiolytic in rodents and primates, J 796

Pharmacol Exp Ther 316(1) (2006) 410-22. 797

[65] E.P. Christian, D.H. Snyder, W. Song, D.A. Gurley, J. Smolka, D.L. Maier, M. Ding, F. 798

Gharahdaghi, X.F. Liu, M. Chopra, M. Ribadeneira, M.J. Chapdelaine, A. Dudley, J.L. Arriza, C. 799

Maciag, M.C. Quirk, J.J. Doherty, EEG-beta/gamma spectral power elevation in rat: a translatable 800

biomarker elicited by GABA(Aalpha2/3)-positive allosteric modulators at nonsedating anxiolytic 801

doses, J Neurophysiol 113(1) (2015) 116-31. 802

[66] B.D. Fischer, J.R. Atack, D.M. Platt, D.S. Reynolds, G.R. Dawson, J.K. Rowlett, Contribution 803

of GABA(A) receptors containing alpha3 subunits to the therapeutic-related and side effects of 804

benzodiazepine-type drugs in monkeys, Psychopharmacology (Berl) 215(2) (2011) 311-9. 805

[67] A.G. de Lucas, P.K. Ahring, J.S. Larsen, I. Rivera-Arconada, J.A. Lopez-Garcia, N.R. Mirza, 806

G. Munro, GABAA alpha5 subunit-containing receptors do not contribute to reversal of 807

inflammatory-induced spinal sensitization as indicated by the unique selectivity profile of the 808

GABAA receptor allosteric modulator NS16085, Biochem Pharmacol 93(3) (2015) 370-9. 809

[68] J. Simon, H. Wakimoto, N. Fujita, M. Lalande, E.A. Barnard, Analysis of the set of GABA(A) 810

receptor genes in the human genome, J Biol Chem 279(40) (2004) 41422-35. 811

[69] S. Zhu, C.M. Noviello, J. Teng, R.M. Walsh, Jr., J.J. Kim, R.E. Hibbs, Structure of a human 812

synaptic GABAA receptor, Nature 559(7712) (2018) 67-72. 813